=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.12 retrieving revision 1.46 diff -u -p -r1.12 -r1.46 --- mandoc/roff.c 2008/11/27 16:54:58 1.12 +++ mandoc/roff.c 2008/12/06 19:41:41 1.46 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.12 2008/11/27 16:54:58 kristaps Exp $ */ +/* $Id: roff.c,v 1.46 2008/12/06 19:41:41 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -16,6 +16,9 @@ * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR * PERFORMANCE OF THIS SOFTWARE. */ +#include +#include + #include #include #include @@ -27,101 +30,52 @@ #include "libmdocml.h" #include "private.h" +#include "roff.h" -/* FIXME: warn if Pp occurs before/after Sh etc. (see mdoc.samples). */ +/* FIXME: First letters of quoted-text interpreted in rofffindtok. */ +/* FIXME: `No' not implemented. */ +/* TODO: warn if Pp occurs before/after Sh etc. (see mdoc.samples). */ +/* TODO: warn about "X section only" macros. */ +/* TODO: warn about empty lists. */ +/* TODO: (warn) some sections need specific elements. */ +/* TODO: (warn) NAME section has particular order. */ +/* TODO: unify empty-content tags a la
. */ +/* TODO: macros with a set number of arguments? */ +/* TODO: validate Dt macro arguments. */ +/* FIXME: Bl -diag supposed to ignore callable children. */ +/* FIXME: Nm has newline when used in NAME section. */ -/* FIXME: warn about "X section only" macros. */ - -/* FIXME: warn about empty lists. */ - -/* FIXME: ; : } ) (etc.) after text macros? */ - -/* FIXME: NAME section needs specific elements. */ - -#define ROFF_MAXARG 32 - -enum roffd { - ROFF_ENTER = 0, - ROFF_EXIT -}; - -enum rofftype { - ROFF_COMMENT, - ROFF_TEXT, - ROFF_LAYOUT, - ROFF_SPECIAL -}; - -#define ROFFCALL_ARGS \ - int tok, struct rofftree *tree, \ - const char *argv[], enum roffd type - -struct rofftree; - -struct rofftok { - int (*cb)(ROFFCALL_ARGS); /* Callback. */ - const int *args; /* Args (or NULL). */ - const int *parents; - const int *children; - int ctx; - enum rofftype type; /* Type of macro. */ - int flags; -#define ROFF_PARSED (1 << 0) /* "Parsed". */ -#define ROFF_CALLABLE (1 << 1) /* "Callable". */ -#define ROFF_QUOTES (1 << 2) /* Quoted args. */ -#define ROFF_SHALLOW (1 << 3) /* Nesting block. */ -}; - -struct roffarg { - int flags; -#define ROFF_VALUE (1 << 0) /* Has a value. */ -}; - struct roffnode { int tok; /* Token id. */ struct roffnode *parent; /* Parent (or NULL). */ - size_t line; /* Parsed at line. */ }; struct rofftree { struct roffnode *last; /* Last parsed node. */ - time_t date; /* `Dd' results. */ - char *cur; + char *cur; /* Line start. */ + struct tm tm; /* `Dd' results. */ + char name[64]; /* `Nm' results. */ char os[64]; /* `Os' results. */ char title[64]; /* `Dt' results. */ char section[64]; /* `Dt' results. */ char volume[64]; /* `Dt' results. */ int state; #define ROFF_PRELUDE (1 << 1) /* In roff prelude. */ - /* FIXME: if we had prev ptrs, this wouldn't be necessary. */ #define ROFF_PRELUDE_Os (1 << 2) /* `Os' is parsed. */ #define ROFF_PRELUDE_Dt (1 << 3) /* `Dt' is parsed. */ #define ROFF_PRELUDE_Dd (1 << 4) /* `Dd' is parsed. */ #define ROFF_BODY (1 << 5) /* In roff body. */ - struct md_mbuf *mbuf; /* Output (or NULL). */ - const struct md_args *args; /* Global args. */ - const struct md_rbuf *rbuf; /* Input. */ - const struct roffcb *cb; + struct roffcb cb; /* Callbacks. */ + void *arg; /* Callbacks' arg. */ }; -static int roff_Dd(ROFFCALL_ARGS); -static int roff_Dt(ROFFCALL_ARGS); -static int roff_Os(ROFFCALL_ARGS); - -static int roff_layout(ROFFCALL_ARGS); -static int roff_text(ROFFCALL_ARGS); -static int roff_comment(ROFFCALL_ARGS); -static int roff_close(ROFFCALL_ARGS); -static int roff_special(ROFFCALL_ARGS); - static struct roffnode *roffnode_new(int, struct rofftree *); -static void roffnode_free(int, struct rofftree *); - +static void roffnode_free(struct rofftree *); static void roff_warn(const struct rofftree *, const char *, char *, ...); static void roff_err(const struct rofftree *, const char *, char *, ...); - +static int roffpurgepunct(struct rofftree *, char **); static int roffscan(int, const int *); static int rofffindtok(const char *); static int rofffindarg(const char *); @@ -130,304 +84,148 @@ static int roffargs(const struct rofftree *, int, char *, char **); static int roffargok(int, int); static int roffnextopt(const struct rofftree *, - int, const char ***, char **); -static int roffparse(struct rofftree *, char *, size_t); -static int textparse(const struct rofftree *, - const char *, size_t); + int, char ***, char **); +static int roffparseopts(struct rofftree *, int, + char ***, int *, char **); +static int roffcall(struct rofftree *, int, char **); +static int roffparse(struct rofftree *, char *); +static int textparse(struct rofftree *, char *); +static int roffdata(struct rofftree *, int, char *); +static int roffspecial(struct rofftree *, int, + const char *, size_t, char **); +static int roffsetname(struct rofftree *, char **); +#ifdef __linux__ +extern size_t strlcat(char *, const char *, size_t); +extern size_t strlcpy(char *, const char *, size_t); +extern int vsnprintf(char *, size_t, + const char *, va_list); +extern char *strptime(const char *, const char *, + struct tm *); +#endif -static const int roffarg_An[] = { ROFF_Split, ROFF_Nosplit, - ROFF_ARGMAX }; -static const int roffarg_Bd[] = { ROFF_Ragged, ROFF_Unfilled, - ROFF_Literal, ROFF_File, ROFF_Offset, ROFF_Filled, - ROFF_Compact, ROFF_ARGMAX }; -static const int roffarg_Bk[] = { ROFF_Words, ROFF_ARGMAX }; -static const int roffarg_Ex[] = { ROFF_Std, ROFF_ARGMAX }; -static const int roffarg_Rv[] = { ROFF_Std, ROFF_ARGMAX }; -static const int roffarg_Bl[] = { ROFF_Bullet, ROFF_Dash, - ROFF_Hyphen, ROFF_Item, ROFF_Enum, ROFF_Tag, ROFF_Diag, - ROFF_Hang, ROFF_Ohang, ROFF_Inset, ROFF_Column, ROFF_Offset, - ROFF_Width, ROFF_Compact, ROFF_ARGMAX }; -static const int roffarg_St[] = { - ROFF_p1003_1_88, ROFF_p1003_1_90, ROFF_p1003_1_96, - ROFF_p1003_1_2001, ROFF_p1003_1_2004, ROFF_p1003_1, - ROFF_p1003_1b, ROFF_p1003_1b_93, ROFF_p1003_1c_95, - ROFF_p1003_1g_2000, ROFF_p1003_2_92, ROFF_p1387_2_95, - ROFF_p1003_2, ROFF_p1387_2, ROFF_isoC_90, ROFF_isoC_amd1, - ROFF_isoC_tcor1, ROFF_isoC_tcor2, ROFF_isoC_99, ROFF_ansiC, - ROFF_ansiC_89, ROFF_ansiC_99, ROFF_ieee754, ROFF_iso8802_3, - ROFF_xpg3, ROFF_xpg4, ROFF_xpg4_2, ROFF_xpg4_3, ROFF_xbd5, - ROFF_xcu5, ROFF_xsh5, ROFF_xns5, ROFF_xns5_2d2_0, - ROFF_xcurses4_2, ROFF_susv2, ROFF_susv3, ROFF_svid4, - ROFF_ARGMAX }; +int +roff_free(struct rofftree *tree, int flush) +{ + int error, t; + struct roffnode *n; -static const int roffchild_Bl[] = { ROFF_It, ROFF_El, ROFF_MAX }; -static const int roffchild_Fo[] = { ROFF_Fa, ROFF_Fc, ROFF_MAX }; -static const int roffchild_Oo[] = { ROFF_Op, ROFF_Oc, ROFF_MAX }; -static const int roffchild_Rs[] = { ROFF_Re, ROFF__A, ROFF__B, - ROFF__D, ROFF__I, ROFF__J, ROFF__N, ROFF__O, ROFF__P, - ROFF__R, ROFF__T, ROFF__V, ROFF_MAX }; + error = 0; -static const int roffparent_El[] = { ROFF_Bl, ROFF_It, ROFF_MAX }; -static const int roffparent_Fc[] = { ROFF_Fo, ROFF_Fa, ROFF_MAX }; -static const int roffparent_Oc[] = { ROFF_Oo, ROFF_Oc, ROFF_MAX }; -static const int roffparent_It[] = { ROFF_Bl, ROFF_It, ROFF_MAX }; -static const int roffparent_Re[] = { ROFF_Rs, ROFF_MAX }; + if ( ! flush) + goto end; -/* Table of all known tokens. */ -static const struct rofftok tokens[ROFF_MAX] = - { - {roff_comment, NULL, NULL, NULL, 0, ROFF_COMMENT, 0 }, /* \" */ - { roff_Dd, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dd */ - { roff_Dt, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Dt */ - { roff_Os, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Os */ - { roff_layout, NULL, NULL, NULL, ROFF_Sh, ROFF_LAYOUT, ROFF_PARSED }, /* Sh */ - { roff_layout, NULL, NULL, NULL, ROFF_Ss, ROFF_LAYOUT, ROFF_PARSED }, /* Ss */ - { roff_text, NULL, NULL, NULL, ROFF_Pp, ROFF_TEXT, 0 }, /* Pp */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* D1 */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Dl */ - { roff_layout, roffarg_Bd, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bd */ - { roff_close, NULL, NULL, NULL, ROFF_Bd, ROFF_LAYOUT, 0 }, /* Ed */ - { roff_layout, roffarg_Bl, NULL, roffchild_Bl, 0, ROFF_LAYOUT, 0 }, /* Bl */ - { roff_close, NULL, roffparent_El, NULL, ROFF_Bl, ROFF_LAYOUT, 0 }, /* El */ - { roff_layout, NULL, roffparent_It, NULL, ROFF_It, ROFF_LAYOUT, ROFF_SHALLOW }, /* It */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ad */ - { roff_text, roffarg_An, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* An */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ar */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Cd */ /* XXX man.4 only */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Cm */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dv */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Er */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ev */ /* XXX needs arg */ - { roff_text, roffarg_Ex, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ex */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fa */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Fd */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fl */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Fn */ /* XXX needs arg */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ft */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ic */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* In */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Li */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_QUOTES }, /* Nd */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Nm */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Op */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ot */ /* XXX deprecated */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pa */ - { roff_text, roffarg_Rv, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Rv */ - { roff_text, roffarg_St, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* St */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Va */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Vt */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xr */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %A */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %B */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %D */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %I */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE}, /* %J */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %N */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %O */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %P */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %R */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* %T */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* %V */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ac */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ao */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Aq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* At */ /* XXX at most 2 args */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bc */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Bf */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bo */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Bq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bsx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Bx */ - {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Db */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Do */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Dq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ec */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, 0 }, /* Ef */ /* FIXME */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Em */ /* XXX needs arg */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Eo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Fx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ms */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* No */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ns */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Nx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ox */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Pf */ - { roff_text, NULL, NULL, NULL, 0, ROFF_LAYOUT, ROFF_PARSED | ROFF_CALLABLE }, /* Po */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Pq */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Ql */ - { roff_layout, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qo */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Qq */ - { roff_close, NULL, roffparent_Re, NULL, ROFF_Rs, ROFF_LAYOUT, 0 }, /* Re */ - { roff_layout, NULL, NULL, roffchild_Rs, 0, ROFF_LAYOUT, 0 }, /* Rs */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sc */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* So */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sq */ - {roff_special, NULL, NULL, NULL, 0, ROFF_SPECIAL, 0 }, /* Sm */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sx */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Sy */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Tn */ - { roff_text, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED }, /* Ux */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xc */ - { NULL, NULL, NULL, NULL, 0, ROFF_TEXT, ROFF_PARSED | ROFF_CALLABLE }, /* Xo */ - { roff_layout, NULL, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Fo */ - { roff_close, NULL, roffparent_Fc, NULL, ROFF_Fo, ROFF_LAYOUT, 0 }, /* Fc */ - { roff_layout, NULL, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Oo */ - { roff_close, NULL, roffparent_Oc, NULL, ROFF_Oo, ROFF_LAYOUT, 0 }, /* Oc */ - { roff_layout, roffarg_Bk, NULL, NULL, 0, ROFF_LAYOUT, 0 }, /* Bk */ - { roff_close, NULL, NULL, NULL, ROFF_Bk, ROFF_LAYOUT, 0 }, /* Ek */ - }; + error = 1; -/* Table of all known token arguments. */ -static const int tokenargs[ROFF_ARGMAX] = - { - 0, 0, 0, 0, - 0, ROFF_VALUE, ROFF_VALUE, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - 0, 0, 0, 0, - }; + if (ROFF_PRELUDE & tree->state) { + roff_err(tree, NULL, "prelude never finished"); + goto end; + } -const char *const toknamesp[ROFF_MAX] = - { - "\\\"", "Dd", "Dt", "Os", - "Sh", "Ss", "Pp", "D1", - "Dl", "Bd", "Ed", "Bl", - "El", "It", "Ad", "An", - "Ar", "Cd", "Cm", "Dv", - "Er", "Ev", "Ex", "Fa", - "Fd", "Fl", "Fn", "Ft", - "Ic", "In", "Li", "Nd", - "Nm", "Op", "Ot", "Pa", - "Rv", "St", "Va", "Vt", - "Xr", "\%A", "\%B", "\%D", - "\%I", "\%J", "\%N", "\%O", - "\%P", "\%R", "\%T", "\%V", - "Ac", "Ao", "Aq", "At", - "Bc", "Bf", "Bo", "Bq", - "Bsx", "Bx", "Db", "Dc", - "Do", "Dq", "Ec", "Ef", - "Em", "Eo", "Fx", "Ms", - "No", "Ns", "Nx", "Ox", - "Pc", "Pf", "Po", "Pq", - "Qc", "Ql", "Qo", "Qq", - "Re", "Rs", "Sc", "So", - "Sq", "Sm", "Sx", "Sy", - "Tn", "Ux", "Xc", "Xo", - "Fo", "Fc", "Oo", "Oc", - "Bk", "Ek", - }; + for (n = tree->last; n; n = n->parent) { + if (0 != tokens[n->tok].ctx) + continue; + roff_err(tree, NULL, "closing explicit scope `%s'", + toknames[n->tok]); + goto end; + } -const char *const tokargnamesp[ROFF_ARGMAX] = - { - "split", "nosplit", "ragged", - "unfilled", "literal", "file", - "offset", "bullet", "dash", - "hyphen", "item", "enum", - "tag", "diag", "hang", - "ohang", "inset", "column", - "width", "compact", "std", - "p1003.1-88", "p1003.1-90", "p1003.1-96", - "p1003.1-2001", "p1003.1-2004", "p1003.1", - "p1003.1b", "p1003.1b-93", "p1003.1c-95", - "p1003.1g-2000", "p1003.2-92", "p1387.2-95", - "p1003.2", "p1387.2", "isoC-90", - "isoC-amd1", "isoC-tcor1", "isoC-tcor2", - "isoC-99", "ansiC", "ansiC-89", - "ansiC-99", "ieee754", "iso8802-3", - "xpg3", "xpg4", "xpg4.2", - "xpg4.3", "xbd5", "xcu5", - "xsh5", "xns5", "xns5.2d2.0", - "xcurses4.2", "susv2", "susv3", - "svid4", "filled", "words", - }; + while (tree->last) { + t = tree->last->tok; + if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) + goto end; + } + if ( ! (*tree->cb.rofftail)(tree->arg)) + goto end; -const char *const *toknames = toknamesp; -const char *const *tokargnames = tokargnamesp; + error = 0; +end: -int -roff_free(struct rofftree *tree, int flush) -{ - int error; + while (tree->last) + roffnode_free(tree); - assert(tree->mbuf); - if ( ! flush) - tree->mbuf = NULL; - - /* LINTED */ - while (tree->last) - if ( ! (*tokens[tree->last->tok].cb) - (tree->last->tok, tree, NULL, ROFF_EXIT)) - /* Disallow flushing. */ - tree->mbuf = NULL; - - error = tree->mbuf ? 0 : 1; - - if (tree->mbuf && (ROFF_PRELUDE & tree->state)) { - /*roff_warn(tree, "prelude never finished");*/ - error = 1; - } - free(tree); + return(error ? 0 : 1); } struct rofftree * -roff_alloc(const struct md_args *args, struct md_mbuf *out, - const struct md_rbuf *in, const struct roffcb *cb) +roff_alloc(const struct roffcb *cb, void *args) { struct rofftree *tree; + assert(args); + assert(cb); + if (NULL == (tree = calloc(1, sizeof(struct rofftree)))) err(1, "calloc"); tree->state = ROFF_PRELUDE; - tree->args = args; - tree->mbuf = out; - tree->rbuf = in; - tree->cb = cb; + tree->arg = args; + (void)memcpy(&tree->cb, cb, sizeof(struct roffcb)); + return(tree); } int -roff_engine(struct rofftree *tree, char *buf, size_t sz) +roff_engine(struct rofftree *tree, char *buf) { tree->cur = buf; + assert(buf); - if (0 == sz) { - roff_warn(tree, buf, "blank line"); + if (0 == *buf) { + roff_err(tree, buf, "blank line"); return(0); } else if ('.' != *buf) - return(textparse(tree, buf, sz)); + return(textparse(tree, buf)); - return(roffparse(tree, buf, sz)); + return(roffparse(tree, buf)); } static int -textparse(const struct rofftree *tree, const char *buf, size_t sz) +textparse(struct rofftree *tree, char *buf) { - - /* Print text. */ + char *bufp; + + /* TODO: literal parsing. */ + + if ( ! (ROFF_BODY & tree->state)) { + roff_err(tree, buf, "data not in body"); + return(0); + } + + /* LINTED */ + while (*buf) { + while (*buf && isspace(*buf)) + buf++; + + if (0 == *buf) + break; + + bufp = buf++; + + while (*buf && ! isspace(*buf)) + buf++; + + if (0 != *buf) { + *buf++ = 0; + if ( ! roffdata(tree, 1, bufp)) + return(0); + continue; + } + + if ( ! roffdata(tree, 1, bufp)) + return(0); + break; + } + return(1); } @@ -444,14 +242,21 @@ roffargs(const struct rofftree *tree, p = buf; + /* + * This is an ugly little loop. It parses a line into + * space-delimited tokens. If a quote mark is encountered, a + * token is alloted the entire quoted text. If whitespace is + * escaped, it's included in the prior alloted token. + */ + /* LINTED */ - for (i = 0; *buf && i < ROFF_MAXARG; i++) { + for (i = 0; *buf && i < ROFF_MAXLINEARG; i++) { if ('\"' == *buf) { argv[i] = ++buf; while (*buf && '\"' != *buf) buf++; if (0 == *buf) { - roff_err(tree, p, argv[i], "unclosed " + roff_err(tree, argv[i], "unclosed " "quote in argument " "list for `%s'", toknames[tok]); @@ -459,8 +264,17 @@ roffargs(const struct rofftree *tree, } } else { argv[i] = buf++; - while (*buf && ! isspace(*buf)) - buf++; + while (*buf) { + if ( ! isspace(*buf)) { + buf++; + continue; + } + if (*(buf - 1) == '\\') { + buf++; + continue; + } + break; + } if (0 == *buf) continue; } @@ -468,10 +282,10 @@ roffargs(const struct rofftree *tree, while (*buf && isspace(*buf)) buf++; } - + assert(i > 0); - if (ROFF_MAXARG == i && *buf) { - roff_err(tree, p, p, "too many arguments for `%s'", toknames + if (ROFF_MAXLINEARG == i && *buf) { + roff_err(tree, p, "too many arguments for `%s'", toknames [tok]); return(0); } @@ -481,7 +295,6 @@ roffargs(const struct rofftree *tree, } -/* XXX */ static int roffscan(int tok, const int *tokv) { @@ -498,14 +311,16 @@ roffscan(int tok, const int *tokv) static int -roffparse(struct rofftree *tree, char *buf, size_t sz) +roffparse(struct rofftree *tree, char *buf) { int tok, t; struct roffnode *n; - char *argv[ROFF_MAXARG]; - const char **argvp; + char *argv[ROFF_MAXLINEARG]; + char **argvp; - assert(sz > 0); + if (0 != *buf && 0 != *(buf + 1) && 0 != *(buf + 2)) + if (0 == strncmp(buf, ".\\\"", 3)) + return(1); if (ROFF_MAX == (tok = rofffindtok(buf + 1))) { roff_err(tree, buf + 1, "bogus line macro"); @@ -514,13 +329,13 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) roff_err(tree, buf + 1, "unsupported macro `%s'", toknames[tok]); return(0); - } else if (ROFF_COMMENT == tokens[tok].type) - return(1); - + } + + assert(ROFF___ != tok); if ( ! roffargs(tree, tok, buf, argv)) return(0); - argvp = (const char **)argv; + argvp = (char **)argv; /* * Prelude macros break some assumptions, so branch now. @@ -529,8 +344,7 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) if (ROFF_PRELUDE & tree->state) { assert(NULL == tree->last); return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - } else - assert(tree->last); + } assert(ROFF_BODY & tree->state); @@ -539,19 +353,19 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) * children are satisfied. */ - if ( ! roffscan(tree->last->tok, tokens[tok].parents)) { - warnx("%s: invalid parent `%s' for `%s' (line %zu)", - tree->rbuf->name, - toknames[tree->last->tok], - toknames[tok], tree->rbuf->line); + if (tree->last && ! roffscan + (tree->last->tok, tokens[tok].parents)) { + roff_err(tree, *argvp, "`%s' has invalid parent `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); } - if ( ! roffscan(tok, tokens[tree->last->tok].children)) { - warnx("%s: invalid child `%s' for `%s' (line %zu)", - tree->rbuf->name, toknames[tok], - toknames[tree->last->tok], - tree->rbuf->line); + if (tree->last && ! roffscan + (tok, tokens[tree->last->tok].children)) { + roff_err(tree, *argvp, "`%s' is invalid child of `%s'", + toknames[tok], + toknames[tree->last->tok]); return(0); } @@ -561,11 +375,6 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) if (ROFF_LAYOUT != tokens[tok].type) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); - - /* - * Check our scope rules. - */ - if (0 == tokens[tok].ctx) return((*tokens[tok].cb)(tok, tree, argvp, ROFF_ENTER)); @@ -596,6 +405,11 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) n = NULL; break; } + if (tokens[n->tok].ctx == n->tok) + continue; + roff_err(tree, *argv, "`%s' breaks `%s' scope", + toknames[tok], toknames[n->tok]); + return(0); } /* @@ -629,16 +443,34 @@ roffparse(struct rofftree *tree, char *buf, size_t sz) * In this, the `El' tag closes out the scope of `Bl'. */ - assert(tree->last); assert(tok != tokens[tok].ctx && 0 != tokens[tok].ctx); + /* LINTED */ + for (n = tree->last; n; n = n->parent) + if (n->tok != tokens[tok].ctx) { + if (n->tok == tokens[n->tok].ctx) + continue; + roff_err(tree, *argv, "`%s' breaks `%s' scope", + toknames[tok], toknames[n->tok]); + return(0); + } else + break; + + + if (NULL == n) { + roff_err(tree, *argv, "`%s' has no starting tag `%s'", + toknames[tok], + toknames[tokens[tok].ctx]); + return(0); + } + + /* LINTED */ do { t = tree->last->tok; if ( ! (*tokens[t].cb)(t, tree, NULL, ROFF_EXIT)) return(0); } while (t != tokens[tok].ctx); - assert(tree->last); return(1); } @@ -664,7 +496,7 @@ static int rofffindtok(const char *buf) { char token[4]; - size_t i; + int i; for (i = 0; *buf && ! isspace(*buf) && i < 3; i++, buf++) token[i] = *buf; @@ -687,6 +519,48 @@ rofffindtok(const char *buf) static int +roffispunct(const char *p) +{ + + if (0 == *p) + return(0); + if (0 != *(p + 1)) + return(0); + + switch (*p) { + case('{'): + /* FALLTHROUGH */ + case('.'): + /* FALLTHROUGH */ + case(','): + /* FALLTHROUGH */ + case(';'): + /* FALLTHROUGH */ + case(':'): + /* FALLTHROUGH */ + case('?'): + /* FALLTHROUGH */ + case('!'): + /* FALLTHROUGH */ + case('('): + /* FALLTHROUGH */ + case(')'): + /* FALLTHROUGH */ + case('['): + /* FALLTHROUGH */ + case(']'): + /* FALLTHROUGH */ + case('}'): + return(1); + default: + break; + } + + return(0); +} + + +static int rofffindcallable(const char *name) { int c; @@ -706,7 +580,6 @@ roffnode_new(int tokid, struct rofftree *tree) if (NULL == (p = malloc(sizeof(struct roffnode)))) err(1, "malloc"); - p->line = tree->rbuf->line; p->tok = tokid; p->parent = tree->last; tree->last = p; @@ -732,12 +605,11 @@ roffargok(int tokid, int argid) static void -roffnode_free(int tokid, struct rofftree *tree) +roffnode_free(struct rofftree *tree) { struct roffnode *p; assert(tree->last); - assert(tree->last->tok == tokid); p = tree->last; tree->last = tree->last->parent; @@ -746,10 +618,102 @@ roffnode_free(int tokid, struct rofftree *tree) static int +roffspecial(struct rofftree *tree, int tok, + const char *start, size_t sz, char **ordp) +{ + + switch (tok) { + case (ROFF_At): + if (0 == sz) + break; + if (0 == strcmp(*ordp, "v6")) + break; + else if (0 == strcmp(*ordp, "v7")) + break; + else if (0 == strcmp(*ordp, "32v")) + break; + else if (0 == strcmp(*ordp, "V.1")) + break; + else if (0 == strcmp(*ordp, "V.4")) + break; + roff_err(tree, start, "invalid `At' arg"); + return(0); + + case (ROFF_Nm): + if (0 == sz) { + if (0 == tree->name[0]) { + roff_err(tree, start, "`Nm' not set"); + return(0); + } + ordp[0] = tree->name; + ordp[1] = NULL; + } else if ( ! roffsetname(tree, ordp)) + return(0); + break; + + case (ROFF_Sx): + /* FALLTHROUGH*/ + case (ROFF_Ex): + if (1 != sz) { + roff_err(tree, start, "`%s' expects one arg", + toknames[tok]); + return(0); + } + break; + + case (ROFF_Sm): + if (1 != sz) { + roff_err(tree, start, "`Sm' expects one arg"); + return(0); + } + + if (0 != strcmp(ordp[0], "on") && + 0 != strcmp(ordp[0], "off")) { + roff_err(tree, start, "`Sm' has invalid argument"); + return(0); + } + break; + + case (ROFF_Ud): + /* FALLTHROUGH */ + case (ROFF_Ux): + /* FALLTHROUGH */ + case (ROFF_Bt): + if (0 != sz) { + roff_err(tree, start, "`%s' expects no args", + toknames[tok]); + return(0); + } + break; + default: + break; + } + + return((*tree->cb.roffspecial) + (tree->arg, tok, tree->cur, ordp)); +} + + +static int +roffcall(struct rofftree *tree, int tok, char **argv) +{ + + if (NULL == tokens[tok].cb) { + roff_err(tree, *argv, "unsupported macro `%s'", + toknames[tok]); + return(0); + } + if ( ! (*tokens[tok].cb)(tok, tree, argv, ROFF_ENTER)) + return(0); + return(1); +} + + +static int roffnextopt(const struct rofftree *tree, int tok, - const char ***in, char **val) + char ***in, char **val) { - const char *arg, **argv; + char *arg, **argv; int v; *val = NULL; @@ -763,7 +727,7 @@ roffnextopt(const struct rofftree *tree, int tok, if (ROFF_ARGMAX == (v = rofffindarg(arg + 1))) { roff_warn(tree, arg, "argument-like parameter `%s' to " - "`%s'", &arg[1], toknames[tok]); + "`%s'", arg, toknames[tok]); return(-1); } @@ -788,10 +752,71 @@ roffnextopt(const struct rofftree *tree, int tok, } +static int +roffpurgepunct(struct rofftree *tree, char **argv) +{ + int i; + + i = 0; + while (argv[i]) + i++; + assert(i > 0); + if ( ! roffispunct(argv[--i])) + return(1); + while (i >= 0 && roffispunct(argv[i])) + i--; + i++; + + /* LINTED */ + while (argv[i]) + if ( ! roffdata(tree, 0, argv[i++])) + return(0); + return(1); +} + + +static int +roffparseopts(struct rofftree *tree, int tok, + char ***args, int *argc, char **argv) +{ + int i, c; + char *v; + + i = 0; + + while (-1 != (c = roffnextopt(tree, tok, args, &v))) { + if (ROFF_ARGMAX == c) + return(0); + + argc[i] = c; + argv[i] = v; + i++; + *args = *args + 1; + } + + argc[i] = ROFF_ARGMAX; + argv[i] = NULL; + return(1); +} + + +static int +roffdata(struct rofftree *tree, int space, char *buf) +{ + + if (0 == *buf) + return(1); + return((*tree->cb.roffdata)(tree->arg, + space != 0, tree->cur, buf)); +} + + /* ARGSUSED */ static int roff_Dd(ROFFCALL_ARGS) { + time_t t; + char *p, buf[32]; if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); @@ -810,11 +835,63 @@ roff_Dd(ROFFCALL_ARGS) return(0); } - /* TODO: parse date. */ - assert(NULL == tree->last); - tree->state |= ROFF_PRELUDE_Dd; + argv++; + + if (0 == strcmp(*argv, "$Mdocdate: December 6 2008 $")) { + t = time(NULL); + if (NULL == localtime_r(&t, &tree->tm)) + err(1, "localtime_r"); + tree->state |= ROFF_PRELUDE_Dd; + return(1); + } + + /* Build this from Mdocdate or raw date. */ + + buf[0] = 0; + p = *argv; + + if (0 != strcmp(*argv, "$Mdocdate:")) { + while (*argv) { + if (strlcat(buf, *argv++, sizeof(buf)) + < sizeof(buf)) + continue; + roff_err(tree, p, "bad `Dd' date"); + return(0); + } + if (strptime(buf, "%b%d,%Y", &tree->tm)) { + tree->state |= ROFF_PRELUDE_Dd; + return(1); + } + roff_err(tree, *argv, "bad `Dd' date"); + return(0); + } + + argv++; + while (*argv && **argv != '$') { + if (strlcat(buf, *argv++, sizeof(buf)) + >= sizeof(buf)) { + roff_err(tree, p, "bad `Dd' Mdocdate"); + return(0); + } + if (strlcat(buf, " ", sizeof(buf)) + >= sizeof(buf)) { + roff_err(tree, p, "bad `Dd' Mdocdate"); + return(0); + } + } + if (NULL == *argv) { + roff_err(tree, p, "bad `Dd' Mdocdate"); + return(0); + } + + if (NULL == strptime(buf, "%b %d %Y", &tree->tm)) { + roff_err(tree, *argv, "bad `Dd' Mdocdate"); + return(0); + } + + tree->state |= ROFF_PRELUDE_Dd; return(1); } @@ -841,8 +918,35 @@ roff_Dt(ROFFCALL_ARGS) return(0); } - /* TODO: parse date. */ + argv++; + if (NULL == *argv) { + roff_err(tree, *argv, "`Dt' needs document title"); + return(0); + } else if (strlcpy(tree->title, *argv, sizeof(tree->title)) + >= sizeof(tree->title)) { + roff_err(tree, *argv, "`Dt' document title too long"); + return(0); + } + argv++; + if (NULL == *argv) { + roff_err(tree, *argv, "`Dt' needs section"); + return(0); + } else if (strlcpy(tree->section, *argv, sizeof(tree->section)) + >= sizeof(tree->section)) { + roff_err(tree, *argv, "`Dt' section too long"); + return(0); + } + + argv++; + if (NULL == *argv) { + tree->volume[0] = 0; + } else if (strlcpy(tree->volume, *argv, sizeof(tree->volume)) + >= sizeof(tree->volume)) { + roff_err(tree, *argv, "`Dt' volume too long"); + return(0); + } + assert(NULL == tree->last); tree->state |= ROFF_PRELUDE_Dt; @@ -850,15 +954,82 @@ roff_Dt(ROFFCALL_ARGS) } +static int +roffsetname(struct rofftree *tree, char **ordp) +{ + + assert(*ordp); + + /* FIXME: not all sections can set this. */ + + if (NULL != *(ordp + 1)) { + roff_err(tree, *ordp, "too many `Nm' args"); + return(0); + } + + if (strlcpy(tree->name, *ordp, sizeof(tree->name)) + >= sizeof(tree->name)) { + roff_err(tree, *ordp, "`Nm' arg too long"); + return(0); + } + + return(1); +} + + /* ARGSUSED */ static int +roff_Ns(ROFFCALL_ARGS) +{ + int j, c, first; + char *morep[1]; + + first = (*argv++ == tree->cur); + morep[0] = NULL; + + if ( ! roffspecial(tree, tok, *argv, 0, morep)) + return(0); + + while (*argv) { + if (ROFF_MAX != (c = rofffindcallable(*argv))) { + if ( ! roffcall(tree, c, argv)) + return(0); + break; + } + + if ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, 1, *argv++)) + return(0); + continue; + } + + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; + + if (argv[j]) { + if ( ! roffdata(tree, 0, *argv++)) + return(0); + continue; + } + + break; + } + + if ( ! first) + return(1); + + return(roffpurgepunct(tree, argv)); +} + + +/* ARGSUSED */ +static int roff_Os(ROFFCALL_ARGS) { + char *p; - if (ROFF_EXIT == type) { - assert(ROFF_PRELUDE_Os & tree->state); - return(roff_layout(tok, tree, argv, type)); - } else if (ROFF_BODY & tree->state) { + if (ROFF_BODY & tree->state) { assert( ! (ROFF_PRELUDE & tree->state)); assert(ROFF_PRELUDE_Os & tree->state); return(roff_text(tok, tree, argv, type)); @@ -871,15 +1042,34 @@ roff_Os(ROFFCALL_ARGS) return(0); } - /* TODO: extract OS. */ + tree->os[0] = 0; + p = *++argv; + + while (*argv) { + if (strlcat(tree->os, *argv++, sizeof(tree->os)) + < sizeof(tree->os)) + continue; + roff_err(tree, p, "`Os' value too long"); + return(0); + } + + if (0 == tree->os[0]) + if (strlcpy(tree->os, "LOCAL", sizeof(tree->os)) + >= sizeof(tree->os)) { + roff_err(tree, p, "`Os' value too long"); + return(0); + } + tree->state |= ROFF_PRELUDE_Os; tree->state &= ~ROFF_PRELUDE; tree->state |= ROFF_BODY; assert(NULL == tree->last); - return(roff_layout(tok, tree, argv, type)); + return((*tree->cb.roffhead)(tree->arg, &tree->tm, + tree->os, tree->title, tree->section, + tree->volume)); } @@ -887,82 +1077,114 @@ roff_Os(ROFFCALL_ARGS) static int roff_layout(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - char *v, *argvp[ROFF_MAXARG]; + int i, c, argcp[ROFF_MAXLINEARG]; + char *argvp[ROFF_MAXLINEARG]; if (ROFF_PRELUDE & tree->state) { - roff_err(tree, *argv, "`%s' disallowed in prelude", + roff_err(tree, *argv, "bad `%s' in prelude", toknames[tok]); return(0); - } - - if (ROFF_EXIT == type) { - roffnode_free(tok, tree); - return((*tree->cb->roffblkout)(tok)); + } else if (ROFF_EXIT == type) { + roffnode_free(tree); + if ( ! (*tree->cb.roffblkbodyout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkout)(tree->arg, tok)); } - i = 0; - argv++; + assert( ! (ROFF_CALLABLE & tokens[tok].flags)); - while (-1 != (c = roffnextopt(tree, tok, &argv, &v))) { - if (ROFF_ARGMAX == c) - return(0); + ++argv; - argcp[i] = c; - argvp[i] = v; - i++; - argv++; - } + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); + if (NULL == roffnode_new(tok, tree)) + return(0); - argcp[i] = ROFF_ARGMAX; - argvp[i] = NULL; + /* + * Layouts have two parts: the layout body and header. The + * layout header is the trailing text of the line macro, while + * the layout body is everything following until termination. + */ - if (NULL == roffnode_new(tok, tree)) + if ( ! (*tree->cb.roffblkin)(tree->arg, tok, argcp, argvp)) return(0); + if (NULL == *argv) + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); - if ( ! (*tree->cb->roffin)(tok, argcp, argvp)) + if ( ! (*tree->cb.roffblkheadin)(tree->arg, tok, argcp, argvp)) return(0); + /* + * If there are no parsable parts, then write remaining tokens + * into the layout header and exit. + */ + if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) + return(0); - /* TODO: print all tokens. */ - - if ( ! ((*tree->cb->roffout)(tok))) + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) return(0); - return((*tree->cb->roffblkin)(tok)); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); } + /* + * Parsable elements may be in the header (or be the header, for + * that matter). Follow the regular parsing rules for these. + */ + + i = 0; while (*argv) { - if (ROFF_MAX != (c = rofffindcallable(*argv))) { - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported " - "macro `%s'", - toknames[c]); + if (ROFF_MAX == (c = rofffindcallable(*argv))) { + assert(tree->arg); + if ( ! roffdata(tree, i++, *argv++)) return(0); - } - if ( ! (*tokens[c].cb)(c, tree, - argv, ROFF_ENTER)) - return(0); + continue; } + if ( ! roffcall(tree, c, argv)) + return(0); + break; + } - /* TODO: print token. */ + /* + * If there's trailing punctuation in the header, then write it + * out now. Here we mimic the behaviour of a line-dominant text + * macro. + */ - argv++; + if (NULL == *argv) { + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); } - if ( ! ((*tree->cb->roffout)(tok))) + /* + * Expensive. Scan to the end of line then work backwards until + * a token isn't punctuation. + */ + + if ( ! roffpurgepunct(tree, argv)) return(0); - return((*tree->cb->roffblkin)(tok)); + if ( ! (*tree->cb.roffblkheadout)(tree->arg, tok)) + return(0); + return((*tree->cb.roffblkbodyin) + (tree->arg, tok, argcp, argvp)); } /* ARGSUSED */ static int -roff_text(ROFFCALL_ARGS) +roff_ordered(ROFFCALL_ARGS) { - int i, c, argcp[ROFF_MAXARG]; - char *v, *argvp[ROFF_MAXARG]; + int i, first, c, argcp[ROFF_MAXLINEARG]; + char *ordp[ROFF_MAXLINEARG], *p, + *argvp[ROFF_MAXLINEARG]; if (ROFF_PRELUDE & tree->state) { roff_err(tree, *argv, "`%s' disallowed in prelude", @@ -970,66 +1192,140 @@ roff_text(ROFFCALL_ARGS) return(0); } + first = (*argv == tree->cur); + p = *argv++; + + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); + + if (NULL == *argv) + return(roffspecial(tree, tok, p, 0, ordp)); + i = 0; - argv++; + while (*argv && i < ROFF_MAXLINEARG) { + c = ROFF_PARSED & tokens[tok].flags ? + rofffindcallable(*argv) : ROFF_MAX; - while (-1 != (c = roffnextopt(tree, tok, &argv, &v))) { - if (ROFF_ARGMAX == c) + if (ROFF_MAX == c && ! roffispunct(*argv)) { + ordp[i++] = *argv++; + continue; + } + ordp[i] = NULL; + + if (ROFF_MAX == c) + break; + + if ( ! roffspecial(tree, tok, p, (size_t)i, ordp)) return(0); - argcp[i] = c; - argvp[i] = v; - i++; - argv++; + return(roffcall(tree, c, argv)); } - argcp[i] = ROFF_ARGMAX; - argvp[i] = NULL; + assert(i != ROFF_MAXLINEARG); + ordp[i] = NULL; - if ( ! (*tree->cb->roffin)(tok, argcp, argvp)) + if ( ! roffspecial(tree, tok, p, (size_t)i, ordp)) return(0); - if ( ! (ROFF_PARSED & tokens[tok].flags)) { + /* FIXME: error if there's stuff after the punctuation. */ - /* TODO: print all tokens. */ + if ( ! first || NULL == *argv) + return(1); - return((*tree->cb->roffout)(tok)); + return(roffpurgepunct(tree, argv)); +} + + +/* ARGSUSED */ +static int +roff_text(ROFFCALL_ARGS) +{ + int i, j, first, c, argcp[ROFF_MAXLINEARG]; + char *argvp[ROFF_MAXLINEARG]; + + if (ROFF_PRELUDE & tree->state) { + roff_err(tree, *argv, "`%s' disallowed in prelude", + toknames[tok]); + return(0); } + first = (*argv == tree->cur); + argv++; + + if ( ! roffparseopts(tree, tok, &argv, argcp, argvp)) + return(0); + if ( ! (*tree->cb.roffin)(tree->arg, tok, argcp, argvp)) + return(0); + if (NULL == *argv) + return((*tree->cb.roffout)(tree->arg, tok)); + + if ( ! (ROFF_PARSED & tokens[tok].flags)) { + i = 0; + while (*argv) + if ( ! roffdata(tree, i++, *argv++)) + return(0); + + return((*tree->cb.roffout)(tree->arg, tok)); + } + + /* + * Deal with punctuation. Ugly. Work ahead until we encounter + * terminating punctuation. If we encounter it and all + * subsequent tokens are punctuation, then stop processing (the + * line-dominant macro will print these tokens after closure). + */ + + i = 0; while (*argv) { if (ROFF_MAX != (c = rofffindcallable(*argv))) { - if (NULL == tokens[c].cb) { - roff_err(tree, *argv, "unsupported " - "macro `%s'", - toknames[c]); + if ( ! (ROFF_LSCOPE & tokens[tok].flags)) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + + if ( ! roffcall(tree, c, argv)) return(0); - } - if ( ! (*tokens[c].cb)(c, tree, - argv, ROFF_ENTER)) + + if (ROFF_LSCOPE & tokens[tok].flags) + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + + break; + } + + if ( ! roffispunct(*argv)) { + if ( ! roffdata(tree, i++, *argv++)) return(0); + continue; } - /* TODO: print token. */ + i = 1; + for (j = 0; argv[j]; j++) + if ( ! roffispunct(argv[j])) + break; - argv++; + if (argv[j]) { + if ( ! roffdata(tree, 0, *argv++)) + return(0); + continue; + } + + if ( ! (*tree->cb.roffout)(tree->arg, tok)) + return(0); + break; } - return((*tree->cb->roffout)(tok)); -} + if (NULL == *argv) + return((*tree->cb.roffout)(tree->arg, tok)); + if ( ! first) + return(1); - -/* ARGSUSED */ -static int -roff_comment(ROFFCALL_ARGS) -{ - - return(1); + return(roffpurgepunct(tree, argv)); } /* ARGSUSED */ static int -roff_close(ROFFCALL_ARGS) +roff_noop(ROFFCALL_ARGS) { return(1); @@ -1038,10 +1334,11 @@ roff_close(ROFFCALL_ARGS) /* ARGSUSED */ static int -roff_special(ROFFCALL_ARGS) +roff_depr(ROFFCALL_ARGS) { - return((*tree->cb->roffspecial)(tok)); + roff_err(tree, *argv, "`%s' is deprecated", toknames[tok]); + return(0); } @@ -1055,8 +1352,8 @@ roff_warn(const struct rofftree *tree, const char *pos (void)vsnprintf(buf, sizeof(buf), fmt, ap); va_end(ap); - (*tree->cb->roffmsg)(tree->args, ROFF_WARN, tree->cur, pos, - tree->rbuf->name, tree->rbuf->line, buf); + (*tree->cb.roffmsg)(tree->arg, + ROFF_WARN, tree->cur, pos, buf); } @@ -1070,6 +1367,7 @@ roff_err(const struct rofftree *tree, const char *pos, (void)vsnprintf(buf, sizeof(buf), fmt, ap); va_end(ap); - (*tree->cb->roffmsg)(tree->args, ROFF_ERROR, tree->cur, pos, - tree->rbuf->name, tree->rbuf->line, buf); + (*tree->cb.roffmsg)(tree->arg, + ROFF_ERROR, tree->cur, pos, buf); } +